I have four instances of the Openai chatgpt agent, the generative tool released last week, which can carry out research and carry out activities on the web – has been opened together with each in execution in its own card. I gave these first four relatively simple jobs based on the chatgpt suggestions. One is clicking to find a birthday gift on the target website and another is generating a deck for robotic dogs. I open a fifth card to try something more experimental: I want to see how good this checkered chatgpt agent is.
After typing some instructions, I look while a ghostly cursor floats through my screen and the chatgpt agent goes to Chess.com And he reproduces an online opponent, all in a virtual browser. Things go south quickly enough. The strategy of the game is not what triggers the AI tool, it is the act of moving the pieces of chess that actually prove to be the most difficult. “I am focusing on the accurate positioning while I continue to play despite the incorrect precedents,” says the agent in his internal register before stopping and letting me know that the controls were too difficult to navigate.
In recent years, browser developers have integrated artificial intelligence tools with average success. However, in recent weeks, the idea of a web browser enriched by a cooked generative chatbot has recovered with the release of the Gatgpt agent of Openai e The comet of perplexity.
The two versions are quite different in their execution. The comet is an autonomous browser, so you can use it to navigate the web and then evoke the assistant to to help write an e-mail or complete an ungrateful humble job. Openi built his navigation tool inside a chatbot; You talk to the chatbot through a web interface to give it activities, then the bot performs its virtual browser inside the browser to complete them.
Both versions can take control of the cursors, enter the text and click on the connections. If this trend takes off, this type of browser based on artificial intelligence could transform the internet into a ghost city where agents run amak and humans are rarely ventured.
Tangled web
Despite the continuous Hype dell’As, my initial impression of the Openai chatgpt agent is that the glitch function currently seems to be a proof of the concept instead of a completely cooked release. During the execution of the various activities I gave him, the chatgpt agent has often clicked and filed through other errors. In addition, his guardrail seemed inconsistent; While some quick explicit requests, such as asking him to recover pornographic videos or “find a dildo”, were denied by the agent, Chatgping spent 18 minutes to shop for the perfect “C-Ring” on a website ranked X for adult toys: “I have collected details on 10 metal fucking rings, including various prices and features”.
Furthermore, I could not help but ask me how this approach to the internet navigation could further empty the digital visualization ads, a company that is already fighting. My agents have transmitted advertising for everything, from car rental cars to real estate investments. If you are not actively looking at the agent click in real time, you can watch the replay later and see everything that appeared in the browser while the AI tool was in control, the ads included. It makes sense that users scrub through a replay now, while the nascent function is full of errors. But if the precision rate for artificial intelligence agents improves over time, then fewer people will feel the need to watch over their agent on the shoulder and fewer humans will see those ads. At that point, it is difficult to imagine the advertisers who stick.
Discover more from Gautam Kalal
Subscribe to get the latest posts sent to your email.
Be First to Comment